If the estimated rate of matching documents is less than global filter limit then

don't build a global filter.

4 files changed, 36 insertions, 1 deletions

diff --git a/searchlib/src/vespa/searchlib/fef/indexproperties.cpp b/searchlib/src/vespa/searchlib/fef/indexproperties.cpp
index fb44b986301..622e437692a 100644
--- a/searchlib/src/vespa/searchlib/fef/indexproperties.cpp
+++ b/searchlib/src/vespa/searchlib/fef/indexproperties.cpp
@@ -290,6 +290,22 @@ NearestNeighborBruteForceLimit::lookup(const Properties &props, double defaultVa
     return lookupDouble(props, NAME, defaultValue);
 }
 
+const vespalib::string GlobalFilterLimit::NAME("vespa.matching.global_filter_limit");
+
+const double GlobalFilterLimit::DEFAULT_VALUE(0.0);
+
+double
+GlobalFilterLimit::lookup(const Properties &props)
+{
+    return lookup(props, DEFAULT_VALUE);
+}
+
+double
+GlobalFilterLimit::lookup(const Properties &props, double defaultValue)
+{
+    return lookupDouble(props, NAME, defaultValue);
+}
+
 } // namespace matching
 
 namespace softtimeout {
diff --git a/searchlib/src/vespa/searchlib/fef/indexproperties.h b/searchlib/src/vespa/searchlib/fef/indexproperties.h
index 30c726caeba..1b4c2e92d8d 100644
--- a/searchlib/src/vespa/searchlib/fef/indexproperties.h
+++ b/searchlib/src/vespa/searchlib/fef/indexproperties.h
@@ -218,6 +218,19 @@ namespace matching {
         static double lookup(const Properties &props);
         static double lookup(const Properties &props, double defaultValue);
     };
+
+    /**
+     * Property to control fallback to not building a global filter
+     * for a query with a blueprint that wants a global filter. If the
+     * estimated ratio of matching documents is less than this limit
+     * then don't build a global filter.
+     **/
+    struct GlobalFilterLimit {
+        static const vespalib::string NAME;
+        static const double DEFAULT_VALUE;
+        static double lookup(const Properties &props);
+        static double lookup(const Properties &props, double defaultValue);
+    };
 }
 
 namespace softtimeout {
diff --git a/searchlib/src/vespa/searchlib/fef/ranksetup.cpp b/searchlib/src/vespa/searchlib/fef/ranksetup.cpp
index e197f095852..249351a4fe5 100644
--- a/searchlib/src/vespa/searchlib/fef/ranksetup.cpp
+++ b/searchlib/src/vespa/searchlib/fef/ranksetup.cpp
@@ -62,7 +62,8 @@ RankSetup::RankSetup(const BlueprintFactory &factory, const IIndexEnvironment &i
       _softTimeoutEnabled(false),
       _softTimeoutTailCost(0.1),
       _softTimeoutFactor(0.5),
-      _nearest_neighbor_brute_force_limit(0.05)
+      _nearest_neighbor_brute_force_limit(0.05),
+      _global_filter_limit(0.0)
 { }
 
 RankSetup::~RankSetup() = default;
@@ -106,6 +107,7 @@ RankSetup::configure()
     setSoftTimeoutTailCost(softtimeout::TailCost::lookup(_indexEnv.getProperties()));
     setSoftTimeoutFactor(softtimeout::Factor::lookup(_indexEnv.getProperties()));
     set_nearest_neighbor_brute_force_limit(matching::NearestNeighborBruteForceLimit::lookup(_indexEnv.getProperties()));
+    set_global_filter_limit(matching::GlobalFilterLimit::lookup(_indexEnv.getProperties()));
 }
 
 void
diff --git a/searchlib/src/vespa/searchlib/fef/ranksetup.h b/searchlib/src/vespa/searchlib/fef/ranksetup.h
index ad793eeaceb..3e127a1e8b5 100644
--- a/searchlib/src/vespa/searchlib/fef/ranksetup.h
+++ b/searchlib/src/vespa/searchlib/fef/ranksetup.h
@@ -60,6 +60,7 @@ private:
     double                   _softTimeoutTailCost;
     double                   _softTimeoutFactor;
     double                   _nearest_neighbor_brute_force_limit;
+    double                   _global_filter_limit;
 
 
 public:
@@ -369,6 +370,9 @@ public:
     void set_nearest_neighbor_brute_force_limit(double v) { _nearest_neighbor_brute_force_limit = v; }
     double get_nearest_neighbor_brute_force_limit() const { return _nearest_neighbor_brute_force_limit; }
 
+    void set_global_filter_limit(double v) { _global_filter_limit = v; }
+    double get_global_filter_limit() const { return _global_filter_limit; }
+
     /**
      * This method may be used to indicate that certain features
      * should be dumped during a full feature dump.